334
May 2019
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
Efficient Processing Software and
Matching Algorithms
Software such as Pix4D, Metashape (formerly
Photoscan), and SimActive CORRELATOR3D
should share the credit for the fast integration of
drones by the mapping industry. Some of these
software packages utilize the SfM approach to
resolve drone imagery geometry. Some utilize
highly advanced matching algorithms, such as
Scale-Invariant Feature Transform, Speeded-
Up Robust Features, and SemiGlobal Matching,
which result in the highest quality tie/pass points
for aerial triangulation and surface models.
Figure 4 illustrates the sheer number of tie/pass
points generated for one UAS-based image. This
density of tie/pass points, with an average of 1,000
to 2,000 points per image, is unprecedented as
conventional photogrammetric practices result in
only up to nine tie/pass points per image (or six
Von Gruber points per stereo model).
Having such high number of tie/pass points
contributes to a high-fidelity bundle block
solution. Figure 5 illustrates the high quality
of digital surface model generated by these new
matching algorithms.
Figure 3. Left: Flight plan of the project Right: 40 images of one ground control point.
2 ASPRS.
/
uploads/2015/01/ASPRS_Positional_Accura-
cy_Standards_Edition1_Version100_Novem-
ber2014.pdf.
3 Abdullah, Q. and C. Kiedrowski. Practical ap-
proach to using the new ASPRS positional accu-
racy standards for digital geospatial data”. In
ASPRS 2019 Annual Conference
, held in Denver,
Colorado, 28 January 2019.
Figure 4. Distribution of matched tie/pass points (represented as orange + symbols) in a
drone image.
“Most traditional photogrammetric software does
not fully utilize the principle of Structure from Motion
(SfM), which is efficient in estimating the geometry of
a block of imagery obtained by drones”
Figure 3 illustrates a ground control point that
appeared in 40 images in a project whose flight
plan is illustrated at the left side of the figure.
Besides strengthening the reliability figure
in the solution, image redundancy provides
the right environment for processing
software that utilizes the concept of multi-ray
photogrammetry, which results in a better aerial
triangulation solution and better products.
